Social stereotypes can be seen as social categories in which we place ourselves and others. While the use of such categories may increase the efficiency of social judgment processes, stereotypes also have well-known deleterious effects, such as when individuals are categorically derogated. In order to understand stereotypes more fully, it's necessary to understand how social groups are encoded and represented in memory. The research proposed is aimed at clarifying and extending existing models for the representation of social categories, with particular attention devoted to the ability of these models to account for perceived differences in group variability. A secondary goal of the proposed research is to develop a more adequate representational account for the out-group homogeneity effect. Four groups of studies are proposed. In the first, three studies are described to study the role of familiarity in the development of social stereotypes and the perception of group variability. These studies involve both experimental manipulations of familiarity and panel surveys to follow the development of group stereotypes over time. The second set of proposed studies focuses on identifying computational procedures that subjects might employ when deriving variability estimates of groups. Existing models of perceived group variability make a variety of assumptions about such computational procedures that these studies will directly examine. The third set of studies explores in-group/out-group differences in how group information is encoded and retrieved. The investigators posit a motivational difference at both encoding and retrieval that may partially account for the out-group homogeneity effect. Finally, the fourth set of studies focuses on the role of the self in forming judgments of groups to which one belongs.